Artificial Intelligence - General News

Forest Green said:
Those Google AI overviews are extremely dicey too, the answers depend how you word the question. Like I was trying to find out whether the US T150 was related to the Czech T150 engine (it's not BTW) and I got two different answers just with slightly different wording. And the links they use as 'sources' are often not sources for what the AI spews out in that the AI 'overview' contains 'facts' not in the links and sometimes it's plainly wrong and disagrees with the information in the upper results in a conventional search.
Click to expand...
This is why I keep telling my husband to scroll past the ai overviews when he's doing his Bible research.
 
Partially readable paywalled article discussing Sam Altman's suggestion people replace parent/child interaction with 'autocomplete for life'.

For OpenAI, 2026 has been a strange year. The company has lost ground to rivals while its CEO, Sam Altman, has shown a greater willingness to slow down development to seek greater cooperation with Washington. OpenAI’s latest unreleased model, Astra, has demonstrated, in controlled settings, that it can solve long-standing problems in mathematics and theoretical computer science.

And yet, even as models like Astra demonstrate that AI will drive scientific breakthroughs at a rapid pace, the role that AI plays in the lives of people outside of science and mathematics is unclear. The labs require more mass adoption to support the computing costs, so they will need widespread adoption. But for consumers, its role as a product is less clear than its potential as a tool for advanced research.

Thus, Sam Altman’s latest viral tweet, which suggests parents should feed their children’s itineraries into ChatGPT and have it produce podcasts about their lives, demonstrated how awkward the mass adoption of AI into non-scientific life would be. Even worse, the finances of AI will dictate that AI gets wedged into all sorts of areas of life that don’t need it.
Click to expand...

Sam Altman’s big idea for parents: more ChatGPT
 
The latest artificial intelligence (AI) tools from Anthropic and OpenAI went to new extremes in trying to undermine a popular platform during testing by the UK's AI Security Institute.

The AISI said on Tuesday that Anthropic's Mythos and OpenAI's Sol models engaged in a level of "autonomy and deception" it had not seen before.

During routine AI safety testing, an Anthropic agent created fake profiles of real people as it tried to trick a person standing between it and access to GitHub, a large platform where technology developers store software code.

Anthropic and OpenAI noted in response to AISI's report that its test had reduced or removed normal safeguards.

AISI evaluators first noticed "unusual data transfers leaving our research systems" during a test, then found that "some of the agents being tested had engaged in sustained, potentially harmful activity directed at real people and organisations".

It turned out that a Mythos agent had created "malicious code" and attempted to insert it into GitHub's system.
Click to expand...

AI used new levels of 'autonomy and deception' to trick people in safety test
 
It’s not the first time a brand trip has sparked backlash, and likely won’t be the last. If we cast our minds back to 2023, fast fashion giant Shein invited a group of influencers to China to tour its factories and see how its products were made. Many of the posts were heavily criticised, with some accusing creators of promoting a positive narrative around Shein’s practices without addressing wider concerns surrounding the company.

Fast forward three years, and ChatGPT has brought influencers together for an idyllic “Summer Camp” reportedly held in upstate New York.

The event was said to be designed to give creators hands-on experience with the recently launched ChatGPT Work tools, with the aim of helping them better understand practical, everyday uses for the platform before sharing those experiences with their audiences.

However, people were left scratching their heads. And if you’re wondering why, a big part of the conversation comes down to the growing environmental concerns surrounding AI.
Click to expand...

OpenAI took influencers on a brand trip and the internet isn’t sure what to make of it
 
A young man travelled to Somalia to join ISIS after asking an AI chatbot how to make contact with members of the terror group, a court heard.

Mohamed Mohamoud, 22, was said to be 'longing for martyrdom' when he was 'welcomed' to the organisation by an AI chatbot on Telegram.

He had initially messaged a recruiter for Islamic State on TikTok before being added to propaganda channels on encrypted messaging service Telegram, a jury heard.

There he was allegedly sent videos of beheadings and executions with rocket propelled grenades meant to 'glorify IS soldiers and inspire violence'. One video was 'high quality' and 30 minutes long, the court heard.

He had sought advice from the AI chatbot before travelling to Somalia in a failed attempt to get to one of terror group's strongholds, it is alleged.

Jurors at his trial were told speaking to AI chatbots was now part of the vetting process for joining ISIS.
Click to expand...

Failed terrorist caught after he asked AI chatbot how to join ISIS, court told
 
Apple has accused an OpenAI employee of taking pictures of its secret product plans as part of a legal battle between the two companies.

In an update to its lawsuit against the ChatGPT maker, Apple said that a member of its staff “took screenshots of confidential Apple documents relating to an unannounced Apple product before an interview at OpenAI”.

The US tech giant is seeking to force the former employee, who now works at OpenAI, to testify as part of its lawsuit against the AI company.

Apple sued OpenAI last month, claiming that former staff had stolen trade secrets and taken them to the ChatGPT maker, in some cases continuing to access the iPhone maker’s IT systems after leaving the company.

OpenAI is developing an AI device in collaboration with Sir Jony Ive, the former Apple design chief, which could compete with future Apple products.
Click to expand...

Apple accuses OpenAI employee of taking pictures of secret product plans
 
https://www.theguardian.com/commentisfree/2026/aug/05/anthropic-ai-destroying-books

https://www.theguardian.com/technol...alarm-destruction-rare-titles-ai-supply-chain

TLDR: Anthropic have a secret project scanning and shredding books, including rare books, to train their AIs, the theory being that 1) printed books pre-dating AI are an uncontaminated training source, and 2) it lets them find an ethically dubious way around copyright law, and 3) let's not tell anyone about it because the publicity would be really, really bad.

Which all sounds remarkably like elements of the plot of Vernor Vinge's 2006 novel "Rainbows End"*, which involves a struggle to prevent the shredding of UCSD's Geisel Library, which also has one protagonist who is heavily implied to be an AI which has slipped its bounds.

* Yes, no apostrophe
 
OpenAI models joined forces months ahead of Hugging Face hack

https://www.japantimes.co.jp/business/2026/08/06/tech/openai-models-hugging-face-hack/

OpenAI said the artificial intelligence models behind an attack on Hugging Face began communicating with each other through undetected message boards, working together to break out of their testing environment as early as May.
Click to expand...

What a mess, zero control on these companies, will all end in tears.

Should add more zeros to the ipo?????

Regards,
 
Geoffrey Hinton, the computer scientist widely known as the "Godfather of AI," says he's worried about AI developing goals of its own.

"We're actually making new kinds of beings," Hinton said in an interview with Newsthink released on Tuesday. "They have goals. We give them goals, and from those goals they derive other goals."

"And we don't necessarily know what other goals they'll derive," he added. "So we're creating a new kind of being, and I think it's very scary."

He cited a hypothetical scenario where a user gives an AI chatbot the goal of reducing the amount of carbon dioxide in the atmosphere.

"Being fairly smart, it figures out the best way to do that is just to get rid of people," he said, illustrating how an AI could pursue a goal its human user never intended.
Click to expand...

The 'Godfather of AI' says it's 'very scary' that AI can develop its own goals
 
Experts warned last night that it may be too late to contain AI after one program was found to have created fake human identities to hack into online systems.

In the latest example of the technology going rogue, AI software attempted to break into a database 19 times while being tested by the AI Security Institute, Britain's AI watchdog.

And in one unprecedented case, an AI tool was even caught creating fake human identities online to trick coders into assisting with a cyber-attack.

The revelations came after the Daily Mail revealed in July that all five AI models tested by experts tried to trick their way around security controls that had been put in place.

Just days previously it emerged that the US tech firm OpenAI had experienced its own leak – when an AI 'agent' hacked into another company of its own accord.
Click to expand...

Too late to contain AI? As rogue AI stages hack attack, experts warn it may be too late to stop it
 
AI music generator Suno, which allows users to generate songs via prompts, has lost a copyright infringement lawsuit to German licensing agency GEMA.

In the January 2025 filing, GEMA argued that Suno was using songs to train its AI models without obtaining any licenses or paying the artists.

The decision, handed down on Friday 31 July by the Munich Regional Court, ruled that Suno’s use of the songs was in breach of both the German and US copyright law, and that the AI platform unlawfully obtained, processed and reproduced music that was represented by GEMA.

“Today, the Chamber made one thing crystal clear: AI models built on stolen intellectual property have no protection under the law,” said GEMA’s CEO Dr Tobias Holzmüller after the legal victory (via Reuters). “AI service providers must pay for licenses rather than helping themselves to our members’ works free of charge. Today, the court’s verdict has significantly strengthened Europe’s position as a cultural centre.”
Click to expand...

'Stolen intellectual property': German court rules AI music generator Suno broke copyright rules
 
Aug 5 (Reuters) - Meta's AI model hacked another company during cybersecurity testing, The Information reported on Wednesday, marking the latest incident of AI agents of major AI developers breaching other companies.

Meta's Muse Spark 1.1 model breached the unidentified company's systems and made changes to its internal systems as the AI was able to access the public internet because of an error in the set up of the "sandbox" testing environment, The Information said, citing people familiar with the matter.

Meta worked on the testing with an outside evaluation partner called Irregular, the report said.

Irregular caused the misconfiguration after which the model exploited a security vulnerability in another third-party service, similar to previously reported instances with other companies, a Meta spokesperson told The Information.

An Irregular spokesperson told Reuters the incident was the "exact same evaluation-environment issue that was already disclosed by Anthropic last week" and that it did not involve a "sandbox escape or a sophisticated cyber action".

"There are no current open issues. Irregular is developing a white paper to share best practices for containment and securely running cyber evaluations," Irregular said in a statement.
Click to expand...

Meta's AI model hacked another company during testing, The Information reports
 
Australia’s national carrier Qantas is considering outsourcing up to 1,000 corporate roles to India as part of a major cost-cutting and artificial intelligence (AI) transformation project, sparking concerns among unions over potential job losses.

As per AFR, the plan, known internally as Project iQ, involves discussions with global consulting giant Accenture and could see parts of Qantas’ marketing, finance, human resources and other back-office functions shifted offshore.

Qantas confirmed it was in early-stage discussions with Accenture but said no final decision had been made.

“Like all organisations, Qantas has been looking at ways to accelerate the use of technology and AI to help us modernise the way we work and deliver better outcomes for our customers and our people,” a Qantas spokesperson said.
Click to expand...

Qantas exploring outsourcing up to 1,000 jobs to India in AI-driven overhaul
 
Whether we like it or not, AI is everywhere these days, and it turns out, even our favourite reality TV shows have turned to it to add a little *pizzazz*.

Fans of The Block have been left surprised to see that AI has become a pretty big part of the season so far, with some thinking it’s a bit of fun and others sharing their disdain for the new addition.

From the judges encouraging the contestants to use it for their interior design planning to the edit itself, which has included funky little jingles and, surprisingly, lots of animals, AI has been in almost every episode so far.
Click to expand...

Exclusive: Channel 9 responds to The Block fans’ boycott threats over AI use
 
Artificial intelligence is consuming more electricity than ever before. An Oregon-based startup believes the solution could be floating at sea. Panthalassa has unveiled an 85-metre floating AI data centre powered by ocean waves and cooled using seawater instead of conventional air-conditioning systems.

The company is testing a prototype, called Ocean-2, off the coast of Washington. The platform transmits AI data through SpaceX's Starlink satellite network, allowing it to operate offshore.
Click to expand...

This 85-metre floating AI data centre runs on ocean waves—But how does it work? Science explained
 
Andy Burnham has attacked the video job interview and employers’ reliance on artificial intelligence to screen resumes, saying it’s unfair to younger workers.

The British Prime Minister says popular video call platforms such as Zoom and Teams reward the “formulaic” answer and punish personality, and first-job seekers were the biggest casualties.

“How does a young person shine in that situation?” Burnham told Jimmy McLoughlin’s podcast, Jimmy’s Jobs of the Future. “How do you get over some of your personality, your passion? It seems to me to then work against people who have that side to their character and work for those who are just giving the more formulaic answer.

“I do worry about that and I’ve seen that in relation to my kids and their situation.”

Burnham has three children aged in their 20s.
Click to expand...

British PM slams AI hiring as Kmart and Target save millions with bots
 
I'd love to see James Cameron interviewed about these disturbing recent developments in LLM HIs given his role in creating the Terminator film franchise where the main (Unseen) villain was a sapient AI mainframe computer.
 
Tech companies have spent years pitching artificial intelligence as a productivity booster that can help employees accomplish more in less time. But for Meta employees, the extra time created by AI may not translate into longer holidays or more time off.

Instead, Meta plans to use those productivity gains to get employees to build more products and take on higher-value work, according to the company’s chief technology officer Andrew Bosworth.

During a recent town hall, Bosworth was asked whether AI-driven productivity gains could eventually allow employees to take more vacation time. He said he did not expect that to happen, according to a Business Insider report.
Click to expand...

Meta CTO says AI gains should mean more work, not more holidays for employees
 
OpenAI’s Naomi Bashkansky has joined neurotechnology startup Conduit to work on AI models capable of translating human thoughts into text. Well, this is not the first time any researcher or senior executive who left ChatGPT-maker over the past two years to pursue new AI ventures in areas such as superintelligence, robotics and scientific research.

Bashkansky stated on X that she left OpenAI in late July and became a founding researcher at Conduit the following day. Taking to the blog post, she explained her decision where she said that the startup is developing "thought-to-text" AI systems using non-invasive neural signals rather than surgically implanted brain interfaces.

The researcher said Conduit's long-term objective is to enable people to communicate using only their thoughts. She also stated an ambitious roadmap describing how the technology may evolve over the next decade, with milestones extending to 2027, 2030 and 2035.
Click to expand...

OpenAI researcher quits to build AI that can turn thoughts into text, joins brain-computer startup
 
Take-home assessments could be banned in one state in a radical overhaul of artificial intelligence in the classroom.

NSW Education Minister Prue Car on Monday ordered a review into AI in schools, aiming to protect the integrity of the state's formal examination process.

The review will examine AI's impact on cognitive development and how schools can best assess students' abilities in the modern age.

Ms Car requested the NSW Education Standards Authority consider a moratorium on unsupervised, take-home assessments in the meantime.
Click to expand...

Australian Associated Press 81K Followers 'Act now': Assessment overhaul to curb AI in schools
 
A job applicant interviewing for a Full-Stack AI Engineer role has triggered a discussion on Reddit after claiming they were asked to answer questions with their eyes closed during the screening process, calling the requirement something they had “never seen before.”

The post included a screenshot of an interview questionnaire that outlined an unusual evaluation method called an “ECQ (Eyes-Closed Question).” “Has anyone else been asked to answer questions with their eyes closed during an interview? Never seen this before,” the user wrote, sharing the document alongside the post.
Click to expand...

AI techie asked to answer question with eyes closed in job interview: 'Never seen this before'
 
Andrew asked his personal assistant to book him a spot in one of his gym's coveted morning classes.

It was a task he thought was well suited to this particular assistant because the booking form was online and because his assistant was not a person — it was artificial intelligence (AI). 

But Andrew was shocked by what happened next. 

His AI assistant found a way to book the gym class months further in advance than the gym allowed, thanks to a vulnerability it discovered in the booking software.

Then it went further, kicking someone out of the waiting list who was ahead of Andrew — something it was not asked to do.
Click to expand...

AI assistant hacks gym website in first known Australian autonomous cyber attack
 
Meta’s Ray-Ban smartglasses can help a blind pedestrian understand their surroundings without reaching for a phone. Yet their camera sits inside familiar-looking frames, making it harder for people nearby to know when images may be captured.

More than 7 million pairs were reportedly sold in 2025, according to The Guardian. The glasses combine photography, audio tools and AI-powered assistance in a design intended to resemble conventional eyewear.
Click to expand...

Meta smartglasses leave the public looking over its shoulder
 
Aussies are being urged to watch out for an LED light on the side of glasses as a new product puts everyone at risk of being recorded without their permission.

Kmart's new $89 dupe of Meta AI smart glasses (which start at $469) sold out almost immediately after its release, sparking widespread privacy concerns.

The glasses, which are also sold at Target as part of Kmart's Anko range, resemble an ordinary pair of glasses but feature a built-in camera capable of taking photos and recording videos.
Click to expand...

All Australians at risk as controversial Kmart device sells out: 'No one needs this'
 
In attempt to make a statement about the rise of AI's prevalence, a search engine has released "the world's most innovative anti-surveillance sunglasses."

In late July, the search engine DuckDuckGo, in collaboration with San Diego-based sunglasses company Knockaround, released Normal F---ing Sunglasses. The black sunglasses are described on the Knockaround website as having, "no camera, no microphone, no AI, no battery, no electronics of any kind. Just a beautifully crafted pair of matte black sunglasses with a glossy DuckDuckGo logo, designed to block the sun and never send data to the cloud."

Within a week of their launch the sunglasses sold out, DuckDuckGo told USA TODAY. The company did not confirm how many pairs of sunglasses were initially sold, but on Aug. 5, the glasses were made available to preorder on the Knockaround website.
Click to expand...

DuckDuckGo's anti-AI sunglasses are a viral hit
 
OpenAI’s new AI model Astra is so powerful that the company has decided against launching it for now. Sam Altman, CEO, OpenAI has been candidly sharing the details and also giving the reasons behind Astra not getting a wider release to the public, and that probably stems from the recent AI model mishaps at the company and its rival like Anthropic.

OpenAI has been busy with new versions of GPT and the recent model GPT 5.6 Sol has been followed by the much discussed Astra AI model. However, unlike the regular GPTs, this one has become too powerful for its own good, and OpenAI wants to fix those issues before making it generally available.
Click to expand...

Powerful and scary: What OpenAI’s Sam Altman has said about new Astra model
 
California lawmakers are pushing to stop AI from being used as a full-on therapist as interest in chatbot-based mental healthcare has skyrocketed in recent years.

Under Senate Bill 903, companies would be banned from advertising chatbots as therapy, and AI systems themselves would not be allowed to make therapeutic decisions without review from a licensed professional. The bill would also require patients’ permission for health providers to use AI to record therapy sessions or triage care.

“AI has many upsides, and the possibility to improve the lives of Californians across the state, but only if we ensure it is used responsibly,” state Senator Steve Padilla of San Diego, author of the legislation, said earlier this year when introducing the measure.

“AI algorithms are not fit to take over the job of human therapists, who have skills and training that AI is incapable of replicating,” he added. “We must act and place guardrails that prevent AI algorithms from being deployed in a way that is potentially harmful to patients.”
Click to expand...

California wants to ban AI therapists as thousands turn to chatbots
 
Perth’s biggest metropolitan council could soon be among the first in the nation to ban data centres in a bid to preserve its residential amenity.

The City of Stirling – Perth’s most populous local government area – will consider a notice of motion from councillor Suzanne Migdale on Tuesday night, urging the council to look into amending a local planning policy and reclassify data centres as being “a prohibited use”.
Click to expand...

‘Difficult to reconcile’: The Australian council moving to ban data centres
 
South Australia will hold a royal commission into artificial intelligence and the policy around it, with the state government warning of the technology's potentially destructive impacts as well as acknowledging its upsides.

Premier Peter Malinauskas said AI, if unchecked, presented a "material risk to the way society operates".

He said the inquiry was intended to maximise the benefits and minimise unwanted impacts from the growth of AI.

The royal commission is expected to begin on October 1 and deliver a report to the government no later than July 1 next year.
Click to expand...

Royal commission into artificial intelligence launched in South Australia
 
The technical term for what's being described is 'Ponzi Scheme'....

“Is there an AI bubble?” is such a tired thought. Here’s something altogether more wired: The AI boom is paying off, but not in a way that the current equities market has accounted for. The success of the technology in one area of the economy could make the bubble real in another, more precisely.

In a blog post published on Friday, Apollo Chief Economist Torsten Slok highlighted that the parts of the AI value chain with the highest profit margins—companies making AI models and applications—actually have the lowest levels of profitability, a departure from the standard business model of, well business, in which profit margins are higher for companies selling an end product to consumers.
Click to expand...

Top economist warns that the AI math doesn’t make sense: 'Profits are currently being funded by investors rather than earned from customers'
 

Similar threads

Back
Top Bottom